Autonomous vehicle control strategies for driving safety

Project description

This project aims to devise a tightly coupled solution to autonomous driving control with dynamic obstacle detection and avoidance, which jointly improves safety and efficiency of traffic streams. Vehicle autonomy has several levels starting from driver-assist systems (level 1) to fully autonomous vehicles AVs (level 5). The driving safety of is arguably one of the most overlooked but crucial aspects for successful proliferation of AVs. This project will start with an exploration of existing driver assist systems crucial for vehicle control and stability like the ESP, ABS, TCS, etc. Next, the project will focus on safety of surrounding vehicles by developing a cooperative path-planning and tracking controller. To operate in a public environment with other human-driven vehicles, will require human driver modeling, obstacle detection and path prediction of human-driven vehicles. A distributed model predictive control MPC approach based on Mixed-integer quadratic programming MIQP for optimal trajectory generation is one potential solution in addition to other algorithms. Cooperative behavior will be introduced by broadcasting planned trajectories of connected AVs. The controller will generate steering and wheel torques while incorporating actuator constraints in control law. The algorithms will be tested first on a high-fidelity model developed using different commercial packages. For real-time RT implementation, vehicle dynamics model of intermediate complexity will be investigated that satisfactorily represents actual vehicle and operate in RT. Vehicle trajectories for tracking include longitudinal and lateral positions, velocities, yaw and roll rates and other vehicle dynamics characteristics of interest. A cooperative obstacle avoidance maneuver will be simulated at different speeds followed by field tests. The prospective candidate will develop a scaled vehicle prototype [1-2] followed by a full-scale vehicle like UWA REV [3]. The platform can be used for other research areas like bifurcation analysis [4], driver modeling [5], lap time optimization studies [6], etc.

[1] https://www.donkeycar.com/
[2] https://www.zmp.co.jp/en/products/robocar/robocar-110x
[3] https://therevproject.com/
[4] FD Rossa, et al. (2012) Bifurcation analysis of an automobile model negotiating a curve, Vehicle System Dynamics, 50:10, 1539-1562.
[5] FD Rossa, et al. (2018) Straight ahead running of a nonlinear car and driver model – new nonlinear behaviours highlighted. Vehicle System Dynamics 56:5, 753-768.
[6] B Olaf, et al. (2021) A convex optimization framework for minimum lap time design and control of electric race cars. IEEE Transactions on Vehicular Technology 70.9: 8478-8489.

Outcomes

• An autonomous platform that can be used for further studies in allied areas.
• An evaluation of different vehicle control strategies for autonomous driving.
• Novel path planning and tracking algorithms.
• Extension of the proposed research to new horizons and using the platform to its fullest potential for multidisciplinary studies.

Essential capabilities

Mathematical modeling, knowledge of vehicle systems, coding skills, good academic record and analytical skills

Desireable capabilities

Hands-on experience in student competitions (BAJA, Formula SAE, etc.), Good technical writing skills, Knowledge of commercial software packages, Embedded system design

Expected qualifications (Course/Degrees etc.)

Bachelor’s degree in Mechanical Engineering/Electronic Engineering and allied areas or a Master’s degree with application to vehicle systems/embedded systems and algorithms.

Candidate Discipline

Autonomous driving Vehicle safety Real-time simulation Artificial intelligence.